Site Home   Archive Home   FAQ Home   How to search the Archive   How to Navigate the Archive   
Compare FPGA features and resources   

Threads starting:
1994JulAugSepOctNovDec1994
1995JanFebMarAprMayJunJulAugSepOctNovDec1995
1996JanFebMarAprMayJunJulAugSepOctNovDec1996
1997JanFebMarAprMayJunJulAugSepOctNovDec1997
1998JanFebMarAprMayJunJulAugSepOctNovDec1998
1999JanFebMarAprMayJunJulAugSepOctNovDec1999
2000JanFebMarAprMayJunJulAugSepOctNovDec2000
2001JanFebMarAprMayJunJulAugSepOctNovDec2001
2002JanFebMarAprMayJunJulAugSepOctNovDec2002
2003JanFebMarAprMayJunJulAugSepOctNovDec2003
2004JanFebMarAprMayJunJulAugSepOctNovDec2004
2005JanFebMarAprMayJunJulAugSepOctNovDec2005
2006JanFebMarAprMayJunJulAugSepOctNovDec2006
2007JanFebMarAprMayJunJulAugSepOctNovDec2007
2008JanFebMarAprMayJunJulAugSepOctNovDec2008
2009JanFebMarAprMayJunJulAugSepOctNovDec2009
2010JanFebMarAprMayJunJulAugSepOctNovDec2010
2011JanFebMarAprMayJunJulAugSepOctNovDec2011
2012JanFebMarAprMayJunJulAugSepOctNovDec2012
2013JanFebMarAprMayJunJulAugSepOctNovDec2013
2014JanFebMarAprMayJunJulAugSepOctNovDec2014
2015JanFebMarAprMayJunJulAugSepOctNovDec2015
2016JanFebMarAprMayJunJulAugSepOctNovDec2016
2017JanFebMarAprMayJunJulAugSepOctNovDec2017
2018JanFebMarAprMayJunJulAugSepOctNovDec2018
2019JanFebMarAprMayJunJulAugSepOctNovDec2019
2020JanFebMarAprMay2020

Authors:A B C D E F G H I J K L M N O P Q R S T U V W X Y Z

Custom Search

Messages from 125600

Article: 125600
Subject: Re: Signetics N82F101F
From: Jim Granville <no.spam@designtools.maps.co.nz>
Date: Tue, 30 Oct 2007 11:44:49 +1300
Links: << >>  << T >>  << A >>
> The part was 82S101 not 82F101 -  somewhere in a chain of phone calls
> the S was turned into an F
> 
> Yes I just saw the QPsemi web site - good site to know about
> 
> Anyone know inexpensive testing equiment for PLA's - there's an
> elaborate functional test procedure

  How many test vectors ?

  Are you using these for new designs, or maintenance ?

  We use the EETools device programmers - ChipMAX 2 / TopMAX 2
(Tm2 has some 82S devices, but memories not the PLA, so you
would need to pgm on something else, and choose a CPLD with
DIp28 Vcc/Gnd mapping )

  We've worked with EETools to improve the vector thruput, and
total vector size.
  You can now load 99,999 vectors, and we sent them a 3.5MB
test file :)

  I've never needed quite that much, but figured if it was practical to
generate that many vectors, it should also be practical to test
to a similar limit.

-jg



Article: 125601
Subject: Re: Xilinx, MIG, UCF: timing constraints for DDR2 DRAM
From: "Eric Crabill" <eric.crabill@xilinx.com>
Date: Mon, 29 Oct 2007 16:28:18 -0700
Links: << >>  << T >>  << A >>
Hi Simon,

This document may be of assistance: 
http://www.xilinx.com/bvdocs/appnotes/xapp458.pdf

Eric

"Simon Heinzle" <sheinzle@inf.ethz.ch> wrote in message 
news:47261129$1@news1-rz-ap.ethz.ch...
> Hi FPGA Group!
>
> I'm struggling to get a fast speed (~ 200 MHz) for the DDR2 DRAM 
> interface, generated with the Xilinx Memory Interface Generator. The 
> complete system consists of a PCI interface, an I/O DMA buffer, a burst 
> module bursting from DMA buffer to the DDR2 DRAM interface.
>
> What is the best way to define setup/hold times for the I/O pads (UCF)? 
> (the RAM interface consists of a bi-dir data bus DQ, some output signals 
> e.g. A and the DRAM clocks CK)
> 1. Using the OFFSET = OUT 5 ns AFTER "SYS_CLK_P"? Unfortunately, this does 
> only work using the Input Clock Pin, but it should probably better be in 
> reference to the DRAM clocks CK)
> 2. Using TIMESPEC "TS_DDR_OUT" = FROM FFS TO "DDR_OUT" 5 ns; ? Is probably 
> better, but I'm not exactly sure.
>
> Furthermore, how would you tackle the problem if the timing at the pads 
> cannot be met?
>
> Thanks in advance for helpful answers and pointers in the right direction!
>
> Best regards,
> Simon
>
>
> 



Article: 125602
Subject: Re: builing a SPI interface in vhdl
From: futzy.r@gmail.com
Date: Tue, 30 Oct 2007 05:17:10 -0000
Links: << >>  << T >>  << A >>
On Oct 25, 9:57 am, techG <giuliopul...@gmail.com> wrote:
> HI, I'm new in FPGA, I have to build a SPI interface (in VHDL) to let
> an fpga read and write a flash memory.
> The fpga is a Xilinx Spartan3E, while the memory is an ST M25P16
> (serial I/O).
> Do you know if is there any built vhdl core to start with?
>
> Thanks in advance
> Giulio

You can try www.opencores.org;
The SPI core interface is quite simple to code up as well.
Cheers


Article: 125603
Subject: Re: FPGA Configuration
From: Thomas Stanka <usenet_nospam_valid@stanka-web.de>
Date: Mon, 29 Oct 2007 23:08:03 -0700
Links: << >>  << T >>  << A >>
On 29 Okt., 11:48, Antti <Antti.Luk...@googlemail.com> wrote:
> On 29 Okt., 11:46, vlsi_freak <clickonjas...@gmail.com> wrote:
> >      Its is true that FPGA boots its program from an external PROM.
> > How external eproms are connected to FPGA. Is there any requirement in
> > making PCB with FPGA.
>
>
> 1 yes
> 2 read the documentation
> 3 yes read the documentation


Atti you should be more precise in aswer 1:
1a) Yes (see eg. www.xilinx.com)
1b) No  (see eg. www.actel.com)

bye Thomas


Article: 125604
Subject: FFT for an arbitrary number of points (not power of 2)
From: Patrick Dubois <prdubois@gmail.com>
Date: 29 Oct 2007 23:23:31 -0700
Links: << >>  << T >>  << A >>
Hi,

I'd like to ask experts here about ideas to perform a FFT on an
arbitrary number of points (for real data).

The cores usually found for an FPGA implementation only permit FFTs on
a number of points that is a power of 2. In our particular case
however, we need to be able to do an FFT on a vector of say, 1025
points. Our current algorithm is to zero-pad this vector to 2048
points.

The problem with this approach is that we nearly double the number of
points for downstream processing. This is very problematic at high
datarates.

We have a few ideas but since this seems like a common problem, I'd
like to ask people here for tips.

Thanks!

Patrick


Article: 125605
Subject: IDE to Flash memory
From: Franck <franck.jullien@gmail.com>
Date: Tue, 30 Oct 2007 06:43:01 -0000
Links: << >>  << T >>  << A >>
Hi,

I'm looking for some ideas to make an IDE to NAND or NOR flash memory
interface. Have you ever heard about something about this (ip,
project, chip...) ?

Thanks.
Franck.


Article: 125606
Subject: Re: FPGA Configuration
From: Antti <Antti.Lukats@googlemail.com>
Date: Tue, 30 Oct 2007 07:20:17 -0000
Links: << >>  << T >>  << A >>
On 30 Okt., 07:08, Thomas Stanka <usenet_nospam_va...@stanka-web.de>
wrote:
> On 29 Okt., 11:48, Antti <Antti.Luk...@googlemail.com> wrote:
>
> > On 29 Okt., 11:46, vlsi_freak <clickonjas...@gmail.com> wrote:
> > >      Its is true that FPGA boots its program from an external PROM.
> > > How external eproms are connected to FPGA. Is there any requirement in
> > > making PCB with FPGA.
>
> > 1 yes
> > 2 read the documentation
> > 3 yes read the documentation
>
> Atti you should be more precise in aswer 1:
> 1a) Yes (see eg.www.xilinx.com)
> 1b) No  (see eg.www.actel.com)
>
> bye Thomas

Hi Thomas

you could have been more precise with the spelling too :)

I was wondering when somebody makes this remark, you are right 1) is
YES/NO
YES
* Xilinx all except S3AN
* Altera all except MAX2 (I think its more a FPGA than CPLD)
* Lattice all except XP,XP2,machXO,ispFPGA

NO
* Actel all
* Quicklogic all
* Altera MAX2
* Lattice XP,XP2,machXO,ispFPGA
* Xilinx S3AN

A N T T I









Article: 125607
Subject: Re: IDE to Flash memory
From: Antti <Antti.Lukats@googlemail.com>
Date: Tue, 30 Oct 2007 07:21:55 -0000
Links: << >>  << T >>  << A >>
On 30 Okt., 07:43, Franck <franck.jull...@gmail.com> wrote:
> Hi,
>
> I'm looking for some ideas to make an IDE to NAND or NOR flash memory
> interface. Have you ever heard about something about this (ip,
> project, chip...) ?
>
> Thanks.
> Franck.

make your homework ;)
http://www.sst.com/products.xhtml/nand_flash_storage/85/

Antti


Article: 125608
Subject: Re: registers are not shown in waveform (xilinx microblaze)
From: "Göran Bilski" <goran.bilski@xilinx.com>
Date: Tue, 30 Oct 2007 08:33:23 +0100
Links: << >>  << T >>  << A >>
Hi,

Even if you had the source code it could be difficult to see the actual 
register file since it's implemented using LUTRAMs.
The register file is spread out in 64 primitives.

I usually just look at the trace signals on MicroBlaze.
There are signals for when MicroBlaze is writing a new value to the register 
file (the value, which register...)
There is plenty of information in the trace signals to get a clue on what 
MicroBlaze is doing.
The signals are described in the MicroBlaze reference guide.

I use these signals when I simulate and when I debugging in target using 
ChipScope.

Göran Bilski


"dartanian" <dartanians@gmail.com> wrote in message 
news:1193674722.374345.163400@22g2000hsm.googlegroups.com...
> Hello fpga community
>
> I try to compile a testbench in order to observe the waveforms of the
> registers of microblaze. The problem is that i cannot see any set of
> MicroBlaze register waves (32 assigned mb registers) after the
> generation of the waveform screen (compile, simulate, run.. ).
>
> Initializing a project in EDK, input microblaze and memory type ip's
> and trying to input adress values to be inserted from memory into
> Microblaze processor. I follow the standard procedure.. compiling,
> generating addresses - libraries - netlist, assign drivers, and
> heading to simulation through modelsim. Rest of waveforms are there
> (sets of mb, opb, lmb, debug, bram...). Can anyone provide me some
> help.
>
> Thank you very much in advance.
> 



Article: 125609
Subject: Re: kicad or orcad virtex5 symbol
From: colin <colin_toogood@yahoo.com>
Date: Tue, 30 Oct 2007 01:30:40 -0700
Links: << >>  << T >>  << A >>
On 8 Oct, 16:19, michel.ta...@gmail.com wrote:
> Hi all,
>
> I'm designing a board based on a virtex5 (XC5VLX50). I'm surprised, I
> can't find virtex 5 symbols for using with orcad or kicad.
> I only found a post explaining how to create symbol in orcad using ISE
> pin file. But it requires Orcad Capture, and I plan tu use kicad
> (http://www.lis.inpg.fr/realise_au_lis/kicad/) for schematics, and
> only Cadence Allegro, the Orcad router for PCB.
>
> This is the first time I work with Xilinx, but I know for Altera FPGA,
> the Orcad symbol was freely downloadable..
>
> So if anyone can help me, or have an interesting link..
>
> Thanks by advance,
>
> Best regards, Michel.

I have just created a generic SPARTAN 3e symbol for ORCAD. I decided
which sizes of FPGA I want to be able to support for the footprint I
want, modified the description of the signals which are different
between fpga size ( the different pins are clearly indicated) then
imported the csv into orcad. It took me half an hour. How many hundred
different variations do you think xilinx should provide?

Colin




Article: 125610
Subject: X3100A design with Synplify 8.8 and foundation 1.5 possible?
From: Antti <Antti.Lukats@googlemail.com>
Date: Tue, 30 Oct 2007 09:48:09 -0000
Links: << >>  << T >>  << A >>
Hi

foundation 1.5 mapper terminates on windows XP with GPF fault, i
wonder if it is caused by incompatibility with winXP or maybe the edif
file generated with synplify is somehow bad?

or any other recommendations how to create bit file from VHDL for
XC3100A FPGA

Antti


Article: 125611
Subject: Re: Is it possible to check how cache memories are mapped to FPGA block rams?
From: Wei Wang <camwwang@gmail.com>
Date: Tue, 30 Oct 2007 10:16:54 -0000
Links: << >>  << T >>  << A >>
On Oct 29, 7:42 pm, Eric Smith <e...@brouhaha.com> wrote:
> Wei Wang <camww...@gmail.com> writes:
> > Hello, I implemented an ARM1176JZFS on a Virtex 5 FPGA, but it seems
> > that the
> > cache memories of the processor are not behaving correctly and I
> > have to turn off the caches for application programs
> > to run correctly on the processor. Just wondering whether it is
> > possible to check whether the processor memories have been mapped
> > correctly onto the block rams on FPGA. Thanks, -Wei
>
> Perhaps if you post the actual project (or put a copy on a web site and
> post a URL), some of us can take a look at it and try to help debug
> it.  Otherwise we don't have very much to go on.
>
> Eric

Eric, I appreciate your willingness to dig further to help, but my
question was how I could check the block ram mapping of my design
which I thought it was quite generic, and I would expect answers, such
as, look at somewhere or look for something in the synthesis log file,
or open fpga editor and look for instantiations of primitives. BTW, I
suppose most of us in this group do not work for ourselves, only lazy
university students would post their entire project and let somebody
else do the work for them. Thanks!


Article: 125612
Subject: Re: FFT for an arbitrary number of points (not power of 2)
From: Andy Botterill <andy@plymouth2.demon.co.uk>
Date: Tue, 30 Oct 2007 11:02:46 +0000
Links: << >>  << T >>  << A >>
Patrick Dubois wrote:
> Hi,
> 
> I'd like to ask experts here about ideas to perform a FFT on an
> arbitrary number of points (for real data).
> 
> The cores usually found for an FPGA implementation only permit FFTs on
> a number of points that is a power of 2. In our particular case
> however, we need to be able to do an FFT on a vector of say, 1025
> points. Our current algorithm is to zero-pad this vector to 2048
> points.

A fast fourier transmformation has to be done in samples of 2**n. This 
significantly simplifies the amount of mathematical processing to be done.

A discrete fourier transmform can be of any size that you want. The 
mathematical processing is a lot more complicated. This will take a lot 
of gates.

\why do you need to sample and process different amounts of data?
> 
> The problem with this approach is that we nearly double the number of
> points for downstream processing. This is very problematic at high
> datarates.
> 
> We have a few ideas but since this seems like a common problem, I'd
> like to ask people here for tips.
> 
> Thanks!
> 
> Patrick
> 

Article: 125613
Subject: Re: FPGA vs ASIC
From: Philip Potter <pgp@see.sig.invalid>
Date: Tue, 30 Oct 2007 11:42:09 +0000
Links: << >>  << T >>  << A >>
Andrew FPGA wrote:
>> . It seems that if we scale the
>> results from the quoted paper, a 45nm FPGA will take only a 5x density
>> hit (allowing hard macros) compared to a 90nm ASIC.
> 
> I was not sure how gate size scales with silicon area used so I did
> not do your calculation - but I certainly did not think 2 process
> generations would give a 20x density improvement.  Do you not think a
> 5x density difference is significant?

A 5x density difference is significant but it is less significant than a 
20x density difference. (As mk rightly points out, my 5x figure is based 
on the faulty assumption that the FPGA is 45nm.) The point I was making 
is that the paper you quoted compares FPGAs and ASICs on the same 
process, which is often not the choice which a business will be given.

Gate density, dynamic power, and clock speed are only some of the 
considerations a business will look at when choosing their 
implementation technology. FPGAs have clear advantages over ASICs in 
some areas, but as you have pointed out ASICs have advantages over FPGAs 
in others.

> Also, they compared the FPGA to a standard cell asic. Presumably with
> full custom the difference would be even greater. (and the NRE FPGA
> advantage would be even greater too of course.)

Yes, you're probably correct here.

Article: 125614
Subject: Re: Is it possible to check how cache memories are mapped to FPGA block rams?
From: morphiend <morphiend@gmail.com>
Date: Tue, 30 Oct 2007 12:18:45 -0000
Links: << >>  << T >>  << A >>
On Oct 30, 6:16 am, Wei Wang wrote:
> On Oct 29, 7:42 pm, Eric Smith wrote:
> > Perhaps if you post the actual project (or put a copy on a web site and
> > post a URL), some of us can take a look at it and try to help debug
> > it.  Otherwise we don't have very much to go on.
>
> > Eric
>
<SNIP>
> BTW, I suppose most of us in this group do not work for ourselves, only lazy
> university students would post their entire project and let somebody
> else do the work for them. Thanks!

Now, now. Don't be too hasty. That's what we do with Xilinx all the
time when their software tools don't work "just right" or you hit that
invisible brick wall that only someone with eyes from inside the brick
wall can tell you why the wall exists (read: Webcases).

So long as what you're working on is not proprietary, the nature of
"open source" mindedness is to share with others and gather feedback.
That's what posting your MHS file would provide.


Article: 125615
Subject: Re: FFT for an arbitrary number of points (not power of 2)
From: "RCIngham" <robert.ingham@gmail.com>
Date: Tue, 30 Oct 2007 07:31:27 -0500
Links: << >>  << T >>  << A >>
If you only need the results at a few frequency points use the "Goertzel
algorithm".

There are versions of FFT defined for non-power-of-2. These are generally
known as "prime factor" algorithms. The Winograd transform is related to
these, but all are complicated in control terms, and so are difficult to
programm on DSP microprocessors, let alone FPGAs.

The power-of-2 FFTs are the most widely used for very good reasons...


Article: 125616
Subject: Re: FFT for an arbitrary number of points (not power of 2)
From: Patrick Dubois <prdubois@gmail.com>
Date: Tue, 30 Oct 2007 12:36:50 -0000
Links: << >>  << T >>  << A >>
On Oct 30, 7:02 am, Andy Botterill <a...@plymouth2.demon.co.uk> wrote:
>
> A fast fourier transmformation has to be done in samples of 2**n. This
> significantly simplifies the amount of mathematical processing to be done.
>
> A discrete fourier transmform can be of any size that you want. The
> mathematical processing is a lot more complicated. This will take a lot
> of gates.
>
> \why do you need to sample and process different amounts of data?

Our instrument is a FTS (Fourier Transform Spectrometer). The number
of time-domain samples is directly related to the resolution of the
instrument. Since we want to give flexibility to the customer, we need
to be able to permit a wide range of samples, say 64 to 32k. Our FFT
module therefore needs to be flexible.

We could limit the number of samples to power of 2s, but this is not
very practical. There is quite a large gap between 16k and 32k points.

Basically I'm looking for a solution that will let me use the Xilinx
FFT core that has a power of 2s limitation, to do FFTs on an arbitrary
number of points.

One idea is the chirp z-transform algorithm, but I'm not sure yet if
it would be suitable for a FPGA implementation.

Patrick


Article: 125617
Subject: Re: Is it possible to check how cache memories are mapped to FPGA block rams?
From: "HT-Lab" <hans64@ht-lab.com>
Date: Tue, 30 Oct 2007 12:46:25 GMT
Links: << >>  << T >>  << A >>

"Eric Smith" <eric@brouhaha.com> wrote in message 
news:m3pryxlncm.fsf@donnybrook.brouhaha.com...
> Wei Wang <camwwang@gmail.com> writes:
>> Hello, I implemented an ARM1176JZFS on a Virtex 5 FPGA, but it seems
>> that the
>> cache memories of the processor are not behaving correctly and I
>> have to turn off the caches for application programs
>> to run correctly on the processor. Just wondering whether it is
>> possible to check whether the processor memories have been mapped
>> correctly onto the block rams on FPGA. Thanks, -Wei
>
> Perhaps if you post the actual project (or put a copy on a web site and
> post a URL), some of us can take a look at it and try to help debug
> it.  Otherwise we don't have very much to go on.
>
> Eric

I am not sure that putting his expensive ARM1176 core on the web would be 
appreciated by ARM :-)

Did you write this core yourself? if not then I am sure you can get some 
excellent support from ARM to fix this issue,

Hans
www.ht-lab.com




Article: 125618
Subject: Re: FFT for an arbitrary number of points (not power of 2)
From: "RCIngham" <robert.ingham@gmail.com>
Date: Tue, 30 Oct 2007 08:02:54 -0500
Links: << >>  << T >>  << A >>
By the way, postings to 'comp.dsp' might elicit more useful answers...


Article: 125619
Subject: Re: FFT for an arbitrary number of points (not power of 2)
From: John_H <newsgroup@johnhandwork.com>
Date: Tue, 30 Oct 2007 13:23:53 GMT
Links: << >>  << T >>  << A >>
Patrick Dubois wrote:
> Hi,
> 
> I'd like to ask experts here about ideas to perform a FFT on an
> arbitrary number of points (for real data).
> 
> The cores usually found for an FPGA implementation only permit FFTs on
> a number of points that is a power of 2. In our particular case
> however, we need to be able to do an FFT on a vector of say, 1025
> points. Our current algorithm is to zero-pad this vector to 2048
> points.
> 
> The problem with this approach is that we nearly double the number of
> points for downstream processing. This is very problematic at high
> datarates.
> 
> We have a few ideas but since this seems like a common problem, I'd
> like to ask people here for tips.
> 
> Thanks!
> 
> Patrick

I'd suggest that an arbitrary number - selected at runtime rather than 
by design - isn't practical but designing for 1025 points might be.

I forget if it was LeCroy or Tektronix, but about a decade ago I read 
the whitepaper on their scope's FFT capability.  Scopes also have the 
"limitation" of non-2^n size samples.  Rather than decompose their FFT 
into 2-element butterflies, they decomposed their 2/5/10 multiple 
waveform samples into 2-element and 5-element FFT nodules.  If the idea 
behind the FFT is to reuse the sin/cos values for a given phase delta, 
decomposing into a non-2^n system can work well.  What obviously won't 
work well for this approach is any size with large prime numbers.

Your 1025 point example calls for 5x5x41 which wouldn't pring so much in 
acceleration.

Sorry I don't have an url for the whitepaper - it's been a decade.

- John_H

Article: 125620
Subject: Re: FFT for an arbitrary number of points (not power of 2)
From: Patrick Dubois <prdubois@gmail.com>
Date: Tue, 30 Oct 2007 13:28:14 -0000
Links: << >>  << T >>  << A >>
On 30 oct, 09:02, "RCIngham" <robert.ing...@gmail.com> wrote:
> By the way, postings to 'comp.dsp' might elicit more useful answers...

I thought about it but prefered to try c.a.f first to get more FPGA
centric solutions. I'll try comp.dsp later if needed :)

Thanks,
Patrick


Article: 125621
Subject: Re: FFT for an arbitrary number of points (not power of 2)
From: Patrick Dubois <prdubois@gmail.com>
Date: Tue, 30 Oct 2007 13:38:19 -0000
Links: << >>  << T >>  << A >>
On 30 oct, 09:23, John_H <newsgr...@johnhandwork.com> wrote:
> I'd suggest that an arbitrary number - selected at runtime rather than
> by design - isn't practical but designing for 1025 points might be.
>
> I forget if it was LeCroy or Tektronix, but about a decade ago I read
> the whitepaper on their scope's FFT capability.  Scopes also have the
> "limitation" of non-2^n size samples.  Rather than decompose their FFT
> into 2-element butterflies, they decomposed their 2/5/10 multiple
> waveform samples into 2-element and 5-element FFT nodules.  If the idea
> behind the FFT is to reuse the sin/cos values for a given phase delta,
> decomposing into a non-2^n system can work well.  What obviously won't
> work well for this approach is any size with large prime numbers.
>
> Your 1025 point example calls for 5x5x41 which wouldn't pring so much in
> acceleration.
>
> Sorry I don't have an url for the whitepaper - it's been a decade.
>
> - John_H

Thanks for the idea. But ideally I'd like to limit myself to power of
2 FFTs so that I can use the Xilinx core. The number of points needs
to be flexible from 64 to 32k. Decomposition into smaller chunks seems
like a promising avenue. Although if one number is decomposible into
two power of 2s (e.g. 8192, which is divisable by 64 and 128), then
that number itself will be a power of 2.

I'd be quite happy with an algorithm that increases the number of
points possible (compared to only power of 2s). Being able to do
_every_ possible number of points is probably too strict a
requirement.

Patrick



Article: 125622
Subject: Re: Xilinx, MIG, UCF: timing constraints for DDR2 DRAM
From: "Simon Heinzle" <sheinzle@inf.ethz.ch>
Date: Tue, 30 Oct 2007 14:41:52 +0100
Links: << >>  << T >>  << A >>
Thanks Eric,

I checked the XApp458 and various XApp7** about Virtex 4 DDR2, as we are 
using a Virtex 4 FX. I'm using the DDR2 controller from the Memory Interface 
Generator.

The main problem are the output delays. Input delays seem to be calibrated 
using tap delay lines. I had a working example with output delays between 
4.4ns and 5.1ns (Data, Address, Control, DRAM clock). I extended that 
example (but no changes to the DRAM interface), now the output delays range 
between 4.4 ns (Data, Control, Dram Clock) and 8.9 ns (Address).

What is are the best constraints to define a tight window? (I'm now using 
TIMESPEC ... = FROM FFS TO "OUT" 5.3 ns, which cannot be met)
What is a good strategy to meet those timing constraints?

Thanks in advance,
Simon

"Eric Crabill" <eric.crabill@xilinx.com> wrote in message 
news:fg5qai$nq03@cnn.xilinx.com...
> Hi Simon,
>
> This document may be of assistance: 
> http://www.xilinx.com/bvdocs/appnotes/xapp458.pdf
>
> Eric
>
> "Simon Heinzle" <sheinzle@inf.ethz.ch> wrote in message 
> news:47261129$1@news1-rz-ap.ethz.ch...
>> Hi FPGA Group!
>>
>> I'm struggling to get a fast speed (~ 200 MHz) for the DDR2 DRAM 
>> interface, generated with the Xilinx Memory Interface Generator. The 
>> complete system consists of a PCI interface, an I/O DMA buffer, a burst 
>> module bursting from DMA buffer to the DDR2 DRAM interface.
>>
>> What is the best way to define setup/hold times for the I/O pads (UCF)? 
>> (the RAM interface consists of a bi-dir data bus DQ, some output signals 
>> e.g. A and the DRAM clocks CK)
>> 1. Using the OFFSET = OUT 5 ns AFTER "SYS_CLK_P"? Unfortunately, this 
>> does only work using the Input Clock Pin, but it should probably better 
>> be in reference to the DRAM clocks CK)
>> 2. Using TIMESPEC "TS_DDR_OUT" = FROM FFS TO "DDR_OUT" 5 ns; ? Is 
>> probably better, but I'm not exactly sure.
>>
>> Furthermore, how would you tackle the problem if the timing at the pads 
>> cannot be met?
>>
>> Thanks in advance for helpful answers and pointers in the right 
>> direction!
>>
>> Best regards,
>> Simon
>>
>>
>>
>
> 



Article: 125623
Subject: Re: FPGA vs ASIC
From: Andy <jonesandy@comcast.net>
Date: Tue, 30 Oct 2007 07:10:49 -0700
Links: << >>  << T >>  << A >>
On Oct 29, 12:15 pm, mk <kal*@dspia.*comdelete> wrote:
> On Mon, 29 Oct 2007 17:09:15 +0000, Philip Potter
>
> <p...@see.sig.invalid> wrote:
> >IIRC Xilinx Virtex-5 is on a 45nm process.
>
> No, Virtex-5 is 65nm. No foundry is selling 45nm production wafers
> yet.http://www.xilinx.com/products/silicon_solutions/fpgas/virtex/virtex5...

No FPGA foundry... Intel is selling 45nm quad core processors now.
Historically, processors and FPGAs are leading applications of the
latest, smallest geometry sizes (for general purpose logic, as opposed
to specific things like memory, etc.).

Andy


Article: 125624
Subject: Updating my bookshelf
From: Nicolas Matringe <nic_o_mat@msn.com>
Date: Tue, 30 Oct 2007 08:02:49 -0700
Links: << >>  << T >>  << A >>
Hello all
I am looking for references of books about digital design (ASIC &
FPGA, high speed, processor, signal processing ... )
What are the books you wouldn't work without ?
Thanks in advance
Nicolas




Site Home   Archive Home   FAQ Home   How to search the Archive   How to Navigate the Archive   
Compare FPGA features and resources   

Threads starting:
1994JulAugSepOctNovDec1994
1995JanFebMarAprMayJunJulAugSepOctNovDec1995
1996JanFebMarAprMayJunJulAugSepOctNovDec1996
1997JanFebMarAprMayJunJulAugSepOctNovDec1997
1998JanFebMarAprMayJunJulAugSepOctNovDec1998
1999JanFebMarAprMayJunJulAugSepOctNovDec1999
2000JanFebMarAprMayJunJulAugSepOctNovDec2000
2001JanFebMarAprMayJunJulAugSepOctNovDec2001
2002JanFebMarAprMayJunJulAugSepOctNovDec2002
2003JanFebMarAprMayJunJulAugSepOctNovDec2003
2004JanFebMarAprMayJunJulAugSepOctNovDec2004
2005JanFebMarAprMayJunJulAugSepOctNovDec2005
2006JanFebMarAprMayJunJulAugSepOctNovDec2006
2007JanFebMarAprMayJunJulAugSepOctNovDec2007
2008JanFebMarAprMayJunJulAugSepOctNovDec2008
2009JanFebMarAprMayJunJulAugSepOctNovDec2009
2010JanFebMarAprMayJunJulAugSepOctNovDec2010
2011JanFebMarAprMayJunJulAugSepOctNovDec2011
2012JanFebMarAprMayJunJulAugSepOctNovDec2012
2013JanFebMarAprMayJunJulAugSepOctNovDec2013
2014JanFebMarAprMayJunJulAugSepOctNovDec2014
2015JanFebMarAprMayJunJulAugSepOctNovDec2015
2016JanFebMarAprMayJunJulAugSepOctNovDec2016
2017JanFebMarAprMayJunJulAugSepOctNovDec2017
2018JanFebMarAprMayJunJulAugSepOctNovDec2018
2019JanFebMarAprMayJunJulAugSepOctNovDec2019
2020JanFebMarAprMay2020

Authors:A B C D E F G H I J K L M N O P Q R S T U V W X Y Z

Custom Search